Glossary · Generative AI
What is generative AI?
Definition: Generative AI is artificial intelligence that creates new content, such as text, images or code, based on patterns learned from large volumes of data.
Generative models are trained on vast numbers of examples and learn the statistical relationships between their elements. When given an instruction, known as a prompt, they produce a new response that follows those patterns. Language models are the type most widely used in business, because they work with text, but there are also models that generate images, audio or video.
Its most useful business application is rarely creating original content. It is working with content that already exists: summarising long documents, drafting replies, extracting data from emails and invoices, classifying requests or translating. Built into processes and connected to company systems, it underpins AI agents and intelligent document processing.
A generative model produces plausible text, not necessarily correct text. It can invent facts, which is known as hallucination, and it knows nothing about the company's internal information unless that information is provided. That is why its answers should be grounded in the company's own sources, amounts and dates checked against systems, and care taken over which personal or confidential data it receives and where it is processed.
Example: after each client meeting, a generative model drafts the summary and action list from the notes, and the account manager reviews it before sending it out.